kkirksey@eng.auburn.edu (Kenneth B. Kirksey) (05/29/91)
I'm having a terrible time using the OL flat checkbox widget. I've
included at the bottom the sample program I wrote to try my hand at using
the flat checkbox. When I run it as is, it gives me a bus error. When
I use dbx to trace it, I get the following dump:
warning: core file read error: address not in data space
`FCheckBox`DrawLabel() at 0xf7731774
`FCheckBox`DrawItem() at 0xf7731c10
`Flat`Redisplay() at 0xf7738698
SendExposureEvent() at 0xf76d8d8c
CompressExposures() at 0xf76d8cf8
`Event`DispatchEvent() at 0xf76d88cc
DecideToDispatch() at 0xf76d913c
XtDispatchEvent() at 0xf76d92f4
XtAppMainLoop() at 0xf76d95cc
XtMainLoop() at 0xf76d95ac
main(argc = 1, argv = 0xf7fffc7c), line 81 in
"/home/u2/jstrang/cost/src/FlatTest.c"
(dbx)
What am I doing wrong? This code is pretty much the sample code given on
pages 180-183 of _An Open Look At Unix_. If anyone out there has had this
problem before, or if anyone can tell me what I'm doing wrong, you would
have my undying gratitude.
Thanx in advance,
Ken
--------------Cut Here-------------------------------------------------------
#include <X11/Intrinsic.h>
#include <X11/StringDefs.h>
#include <Xol/OpenLook.h>
#include <Xol/FCheckBox.h>
#include <strings.h>
#include <stdio.h>
static String itemFields[] = {
XtNlabel
};
#define NUM_ITEM_FIELDS 1
typedef struct _Item
{
String label;
} Item;
Arg args[12];
int n;
static Item items[] = {
{"CSE100"},
{"foo"},
{"foo"},
{"foo"},
{"CSE120"},
{"foo"},
{"foo"},
{"foo"},
{"CSE200"},
{"foo"},
{"foo"},
{"foo"},
{"CSE2200"},
{"foo"},
{"foo"},
{"foo"}
};
#define NUM_ITEMS 16
Widget toplevel,
rs_flat_checkbox;
main (argc, argv)
int argc;
char **argv;
{
/*-----------------------------------------------------------------------+
| Initialize Top Level Shell |
+-----------------------------------------------------------------------*/
toplevel = OlInitialize (
"flatTest",
"FlatTest",
NULL,
0,
&argc, argv);
n = 0;
XtSetArg (args[n], XtNlayoutType, OL_FIXEDCOLS); n++;
XtSetArg (args[n], XtNmeasure, 4); n++;
XtSetArg (args[n], XtNitems, items); n++;
XtSetArg (args[n], XtNnumItems, NUM_ITEMS); n++;
XtSetArg (args[n], XtNitemFields, itemFields); n++;
XtSetArg (args[n], XtNnumItemFields, NUM_ITEM_FIELDS); n++;
XtSetArg (args[n], XtNlabelJustify, OL_RIGHT); n++;
rs_flat_checkbox = XtCreateManagedWidget(
"rs_flat_checkbox",
flatCheckBoxWidgetClass,
toplevel,
args, n);
/*-----------------------------------------------------------------------+
| Go To It ............... |
+-----------------------------------------------------------------------*/
XtRealizeWidget (toplevel);
XtMainLoop();
}merlyn@attunix.att.com (Steve Humphrey) (05/30/91)
> I'm having a terrible time using the OL flat checkbox widget. I've > included at the bottom the sample program I wrote to try my hand at using > the flat checkbox. When I run it as is, it gives me a bus error. > I use dbx to trace it, I get the following dump: > warning: core file read error: address not in data space > `FCheckBox`DrawLabel() at 0xf7731774 > ... > > typedef struct _Item > { > String label; > } Item; > > static Item items[] = { > {"CSE100"}, > ... > > n = 0; > XtSetArg (args[n], XtNitems, items); n++; > XtSetArg (args[n], XtNnumItems, NUM_ITEMS); n++; > > rs_flat_checkbox = XtCreateManagedWidget( > "rs_flat_checkbox", > flatCheckBoxWidgetClass, > toplevel, > args, n); > ... I tried your code exactly as sent; it works fine on my '386 running the OPEN LOOK 4.0 toolkit (equivalent to the OpenWindows 2.5 OLIT). I have not tried it yet on a Sun. However, the problem MAY be in the definition of the Item structure. Now the ``Open Look at UNIX'' book has exactly the same problem, and I thought John Miller tested his examples on a Sun, so something else may be wrong. The XtNitems resource takes an array of items, each of which is defined by a structure of XtArgVal's. This parallels the way attributes are passed in regular XtCreateWidget or XtSetValues calls, for exactly the same reason: The widgets (or Intrinsics) do not know a priori the type of each value, thus the values must be coerced to a common type for passing from the client to the widget. So the correct code fragment from the above should be: > typedef struct _Item > { > XtArgVal label; > } Item; > > static Item items[] = { > { (XtArgVal)"CSE100" }, > ... XtArgVal is defined as a "long" on Suns, versus the "char*" definition of String. Now I would think this would be OK, but I'm not familiar with native types of a SPARC chip. Even if it turns out this is not the problem at hand, I suggest using the XtArgVal type in the definition of the XtNitems array's structure. Eventually this *will* save you a core dump. Steve Humphrey UNIX System Laboratories
kkirksey@eng.auburn.edu (Kenneth B. Kirksey) (05/30/91)
In article <1991May29.182035.26184@cbnewsm.att.com> merlyn@attunix.att.com (Steve Humphrey) writes: > >I tried your code exactly as sent; it works fine on my '386 running >the OPEN LOOK 4.0 toolkit (equivalent to the OpenWindows 2.5 OLIT). >I have not tried it yet on a Sun. However, the problem MAY be in the >definition of the Item structure. Now the ``Open Look at UNIX'' book >has exactly the same problem, and I thought John Miller tested his >examples on a Sun, so something else may be wrong. > >The XtNitems resource takes an array of items, each of which is >defined by a structure of XtArgVal's. This parallels the way attributes >are passed in regular XtCreateWidget or XtSetValues calls, for exactly >the same reason: The widgets (or Intrinsics) do not know a priori >the type of each value, thus the values must be coerced to a common >type for passing from the client to the widget. > >So the correct code fragment from the above should be: > >> typedef struct _Item >> { >> XtArgVal label; >> } Item; >> >> static Item items[] = { >> { (XtArgVal)"CSE100" }, >> ... > >XtArgVal is defined as a "long" on Suns, versus the "char*" definition >of String. Now I would think this would be OK, but I'm not familiar >with native types of a SPARC chip. > >Even if it turns out this is not the problem at hand, I suggest >using the XtArgVal type in the definition of the XtNitems array's >structure. Eventually this *will* save you a core dump. > >Steve Humphrey >UNIX System Laboratories Steve, Thanks for the tip, but it didn't work. I still get the bus error in the same place. I don't think I mentioned it in my original post, but I`m running OLIT 2.5 under X11R4 on Sun SparcStations, if it makes any difference. Thanx, Ken +-------------------------+---------------------------------------------+ \ Ken Kirksey \ "Genetic Drifter, Quarantined in my own \ \ Computer Engineering \ dimension. \ \ Auburn University \ Hey Look! It Smokes! It Drinks! \ \ \ It Philosophizes! It's Loving this new \ \ kkirksey@eng.auburn.edu \ Attention!" \ \ AOL: kkirksey1 \ -Steve Taylor \ +--------------------------+---------------------------------------------+
bradst@laguna.ai.sri.com (John Bradstreet) (06/01/91)
In regards to the problem with OL Flat Checkboxes (and flat Exclusives and flat NonExclusives) causing core dumps on sparcs under OLIT 2.5, I believe the problem is inside the Sun code. Luckilly, Sun seems to be aware of this, and seemed to have fixed the problem for the next release. I was experiencing the same problem myself, but I just recompiled and ran the example under 3.0 beta, and it worked. The next question is when 3.0 is coming out for real. --